Upload
truongtuyen
View
224
Download
0
Embed Size (px)
Citation preview
Paper The New Metric and the New
Software Tool for Determining QoS
in the Short Messages Service
in Mobile NetworksMarcus Rompf1 and Tadeus Uhl2
1 Flensburg University of Applied Sciences, Flensburg, Germany2 Maritime University of Szczecin, Szczecin, Poland
Abstract—With traveling professions and nomadic lifestyles
on the increase there is a growing need for comprehensive mo-
bile networking within the Next Generation Network, and the
security, reliability and technical independency of the Short
Message Service (SMS) is becoming more and more indis-
pensable for business processes. Since competitiveness goes
hand-in-hand with this business capability, and methods to
determine the quality of this service are becoming equally in-
dispensable. This paper presents possible methods to eval-
uate the “Completion Rate of SMS” and “End-to-End De-
livery Time” and combinations of them (new metric in this
paper). A software tool, QoSCalc(SMS), has been developed
in collaboration with the authors in order to verify whether
service providers are abiding by the Service Level Agree-
ments. The tool has been tested in a real measurement envi-
ronment and the results are presented graphically here and
interpreted.
Keywords—communication network, ETSI TS 102-250-2, QoS
measurement environment, QoS measurement techniques, queu-
ing model, SMS service.
1. Introduction
3G networks are becoming more and more widespread
every day. Their services also include the Short Message
Service (SMS). Although their popularity has experienced
a slight decrease in private use, the trend in business use
is upward. Constant availability has become a prerequisite
of both traveling professions and mobile lifestyles. Board-
ing passes via SMS, and SMS TAN (Transaction Authen-
tication Number) for online banking have become com-
mon usages and are making SMS increasingly attractive
for competitive businesses for which security, reliability
and technical independency are crucial factors. “Service
Level Agreements” are signed to guarantee a specific end-
to-end quality of service, which can, however, seldom be
verified conclusively by the average service subscriber.
In November 2009 the European Parliament and Euro-
pean Council adopted Directive 2009/136/EC [1] amending
Directive 2002/22/EC [2] on universal service and users’
rights relating to electronic communications networks and
services, Directive 2002/58/EC [3] concerning the pro-
cessing of personal data and the protection of privacy in
the electronic communications sector and Regulation (EC)
No. 2006/2004 [4] on cooperation between national au-
thorities responsible for the enforcement of consumer pro-
tection laws. Within this directive, providers of electronic
communications services that allow calls should not only
ensure that their customers are adequately informed about
the limitations of the services but also about the routing
of emergency calls. Furthermore, information about ser-
vices, which are not provided over a switched telephone
network, should also cover the level of reliability of the
access and provide caller location information in compari-
son with that of a service that is provided over a switched
telephone network, taking into account current technology
and quality standards and any quality of service parameters
specified under Directive 2002/22/EC (Universal Service
Directive) [2].
Especially with regarding to Quality of Service (QoS),
Member States of EU shall ensure that national regulatory
authorities empowered to require service providers to pub-
lish comparable, adequate and up-to-date information for
end-users on the quality of their services, when requested.
These national regulatory authorities may specify, amongst
other things, the QoS parameters to be measured and the
contents, form and manner in which the information is to
be published. In addition, the national regulatory author-
ities should be able to set minimum QoS requirements to
prevent degradation, traffic-slowing or traffic-hindering. As
a result, it can be said that the European Parliament and the
European Council are not only generally aware of QoS and
all that it entails, but are also ensuring that the Member
States apply their directives.
In 2011, the ETSI technical committee Speech and multi-
media Transmission Quality (STQ) produced two multi-part
technical specifications to cover “QoS aspects for popular
services in mobile networks” [5] and “User related QoS
parameter definitions and measurements” [6]. The second
part of [5] in particular discusses the topic “Definition of
Quality of Service parameters and their computation”, and
contains an abstract definition, which gives a generic de-
scription of the parameter, an abstract equation and the
corresponding user and technical trigger points. The sec-
3
Marcus Rompf and Tadeus Uhl
Fig. 1. SMS traffic growth in 1992–2013 and 2015–2017 forecast. (See color pictures online at www.nit.eu/publications/journal-jtit)
ond part of [6] is devoted to the topics “Voice telephony,
Group 3 fax, modem data services, and SMS”.
Research on the current telecommunications market has re-
vealed that SMS still plays an important role even in the age
of over-the-top (OTT) Messaging, which also has become
very popular. In 2011 the world press declared SMS for
“dead” and heralded OTT Messaging as its replacement.
SMS continues to offer a few benefits such as its flexible
platform and reliability. Figure 1 shows the growth of SMS
traffic [7].
It is obvious that SMS was still popular with about 7.6 tril-
lion messages in 2013. Although it has now entered the
decline phase, it is worth taking a closer look at SMS
nonetheless.
The examination will start with a presentation of the QoS
parameters for SMS as they stand in the ETSI Recommen-
dation TS 102-250-2 [6]. Then, a new metric for SMS
will be defined and illustrated. The paper will then turn
to an introduction of the new tool QoSCalc(SMS), begin-
ning with its general architecture and functional principle.
After that, there is a description of how the new tool was
put to the test in a series of practical measurements. The
paper closes with a conclusion and an outlook.
2. QoS Parameters for SMS
This chapter introduces the main QoS parameters recom-
mended in ETSI TS 102-250-2 [6]:
SMS Service Non-Accessibility. The service non-accessi-
bility denotes the probability that the end user cannot access
the SMS when requested while it is offered by display of
the network indicator on the user equipment.
SMS Service Access Delay. The service access delay is the
time period between sending a short message to the network
and receiving a send confirmation from the network at the
origin side.
SMS Successful SMS Ratio. This ratio describes the
probability that a user can send a short message from a ter-
minal to the Short Message Center and is applicable for
service providers offering the SMS. Therefore, it is recom-
mended that result statistics should return the percentage
of successfully sent short messages, the number of obser-
vations and its calculated absolute accuracy limits for 95%
confidence.
Measurements should be scheduled so as to reflect accu-
rately traffic variations over the hours of a day, the days
of the week and the months of the year. The parameter
is intended to measure a combination of network acces-
sibility and congestion in the signalling channels of the
SMS system throughout service operator’s claimed area of
coverage.
Completion Rate for SMS. This describes the ratio of
correctly received short messages to those sent, and is also
applicable to SMS service providers. It is recommended
that the results that provide statistics should return the ratio
of sent and correctly received short messages, the number
of observations used and its calculated absolute accuracy
limits for 95% confidence.
Measurements should be scheduled so as to reflect accu-
rately traffic variations over the hours of a day, the days of
the week and the months of the year.
End-to-End Delivery Time for SMS. This time is the pe-
riod that elapses between the sending of a short message
from the sender’s terminal equipment to a Short Message
Centre and the receiving of that message on the receiver’s
terminal equipment, and is applicable for service operators
providing the SMS. Result-providing statistics should return
the mean value in seconds for sending and receiving short
messages, the time in seconds within which the fastest 95%
4
The New Metric and the New Software Tool for Determining QoS in the Short Messages Service in Mobile Networks
of short messages are sent and received and the number of
observations performed.
Measurements should be scheduled so as to reflect accu-
rately traffic variations over the hours of a day, the days of
the week and the months of the year. The above-mentioned
measurements for estimating QoS parameters should be
done either on real traffic, on test calls for short messages
sent among a representative population of local exchanges
or on a combination of the above.
A measurement system can be designed for these QoS pa-
rameters that takes “Completion Rate for SMS” and the
“End-to-End Delivery Time for SMS” into account.
Research of current QoS measurement techniques for
the SMS has culminated in a large professional solution
called QualiPoc Android, from SwissQual AG, a Rohde
& Schwarz company [8]. An alternative system, QoSCalc
(SMS), is presented in this work. The next section intro-
duces the new metric for evaluating the QoS in SMS.
3. New Metric for QoS in SMS
The conceptual idea is to introduce the SMS Quality Factor
(SMSQ-F), which takes completion rate and end-to-end de-
livery time into account. The formulas of this metric have
the following definition:
SMSQ−F=CRResult ·CRWeighting+DTResult ·DTWeighting , (1)
where:
CRResult =
1
n∑Completicon Rate(n)
Completation Rate Threshold, (2)
DTResult =Delivery Time Threshold
1
n
n
∑n=1
Delivery Time(n)
. (3)
Both thresholds should be chosen judiciously depending on
individual needs. Usually, completion rate threshold is 1,
so that every short message should be successful delivered.
The threshold in seconds for the delivery time will reflect
professional requirements, e.g. alarm systems ≤ 3 or online
banking ≤ 7.
The respective weightings must have the sum of 1 and
should be chosen to reflect how the evaluator gauges their
importance.
The following example is chosen:
• completion rate threshold = 1,
• delivery time threshold = 10 s,
• weighting completion rate = weighting delivery time
= 0.5.
Table 1 gives some example results.
From Table 1 it can be seen that the proposed metric, based
on Eq. (1), works well. Furthermore, it can be seen that
Table 1
Example results of the new metric
n Average CR Average DT [s] CRResult DTResult SMSQ-F
31 1 10 1.00 1.00 1.00
31 1 11 1.00 0.91 0.95
31 1 4 1.00 1.00 1.00
31 1 40 1.00 0.25 0.63
31 1 22 1.00 0.45 0.73
31 0.9 15 0.90 0.67 0.78
31 0.8 16 0.80 0.63 0.71
31 0.7 10 0.70 1.00 0.85
31 0.6 4 0.60 1.00 0.80
31 0.5 19 0.50 0.53 0.51
31 0.4 7 0.40 1.00 0.70
the values of SMSQ-F are within the interval [0, 1]. To be
able to adapt these values to a respective (Quality of Expe-
rience; subjective evaluation) QoE scale, the following rela-
tionship has been chosen, based on the QoS measurement
technique “Apdex-Index” for the WWW service [9]–[10]
(see Fig. 2).
Fig. 2. Translation of QoS values into QoE values for the
SMSQ-F.
In general, the proposed metric can be described as follows:
CRResult =
1 for1
n
n
∑n=1
Completion Rate(n) =
= Completion Threshold
0 for n=0∪Completion Rate Threshold=0
else
1
n
n
∑n=1
Completion Rate(n)
Completion Rate Threshold
,
(4)
5
Marcus Rompf and Tadeus Uhl
DTResult =
1 for1
n
n
∑n=1
Delivery Time(n) =
= Delivery Time Threshold
0 for n=0∪Delivery Time=0
elseDelivery Time Threshold
1
n
n
∑n=1
Delivery Time(n)
. (5)
Furthermore, this metric has been implemented in the
new tool QoSCalc (SMS), neatly customizing it for the
QoS evaluation for SMS. The next chapter introduces this
tool.
4. The New Tool for Analysing QoS
in SMS
The tool that was developed in the course of this work
to analyze QoS in the SMS service has been called
QoSCalc(SMS). It can be used for both objective and sub-
jective analysis of service quality of SMS in mobile net-
works. Therefore, a concept for a new measurement system
has been formulated that takes both the completion rate
for SMS and the end-to-end delivery time for SMS into
account. It is therefore necessary to understand the idea
behind merging these QoS parameters.
Once a short message has been correctly sent and delivered,
the completion rate for SMS provides statistics about the
ratio of correct short messages received to short messages
sent from all observations, while end-to-end delivery time
for SMS provides statistics about the delivery time of short
messages. By merging these QoS parameters it is possible
to provide statistics about the QoE using the SMS.
Fig. 3. Conceptual measurement environment.
In example users will be unsatisfied when a sent short
message may be correctly received but delivered only af-
ter 10 minutes. They might be equally upset by promptly
delivered but error-laden messages. In professional imple-
mentations, time-criticalness is just as important factor as
reliability. When it comes to security and alerting sys-
tems, business implementations stand or fall depending on
time-criticalness and reliability. This concept introduces an
approach to estimate such appropriate QoS parameters.
The general architecture is described in Fig. 3.
The test engineer is responsible for the measurement setup
and controls the test control station, which in turn controls
the SMS send and receive stations. The send station and the
receive station are synchronized and communicate via the
service provider. The chosen hardware platform is the mo-
bile phone HTC Desire 620G dual SIM on Android 4.4.2.
The implementation is based on the block diagram shown
in Fig. 4.
Fig. 4. Conceptual description for the implementation.
The entire solution was written in the programming lan-
guage Java and developed in Android Studio 2.1.2. The
compiled SDK Version is API 23: Android 6.0 and the
Build Tools Version is 23.0.0. The dependencies are set as
follows:
• support: AppCompat 23.1.1 and Design 23.1.1,
• visualization: MPAndroidChartLibrary 2.1.6,
• time synchronization via Network Time Protocol
(NTP): Apache Commons Net 3.4,
• several permissions, such as Receive SMS,
Send SMS and Access Network State, need to
be set.
The basic implemented functionalities are:
• Time synchronization. The time synchronization is
done via the NTP and the server involved is the de-
fault one configured in the system. Then the differ-
ence between the system and server time is calculated
as an offset for the SMS timestamp. Note: it is not
necessary to use the offset correction between send-
ing station and receiving station time synchronization
because the environment does not need to be accurate
to within less than 10 ms.
• Send SMS. Each SMS starts with the string “Mes-
sage”, followed by a zero-initialized, pre-increment
counter and the respective synchronized timestamp.
After 20 short messages have been sent the sending
station goes into the idle state.
6
The New Metric and the New Software Tool for Determining QoS in the Short Messages Service in Mobile Networks
• Receive SMS. Each received SMS is stored as
a structure with the message body, originating ad-
dress, receive time and sending time.
• Analysis. The analysis is done for the “End-to-End
Delivery Time for SMS”, i.e. the difference between
sending and receiving time, the “Completion Rate for
SMS”, i.e. the difference between successful sent and
received short messages, and finally the calculation
of the SMSQ-F.
After being defined, the functionalities have been linked
to the control elements and are ready for the first tests, as
shown in Figs. 5–8.
Fig. 5. Start-up screen QoSCalc(SMS).
Fig. 6. Receive station QoSCalc(SMS) – LTE.
Start-up screen. The Start-up screen allows the user to
choose between sending station and receive station. By de-
fault it is displayed whenever a synchronization error should
occur.
Sending/receive station. The two stations can be time syn-
chronized with the Sync button. Of course, the receive sta-
tion must already have been started. After a measurement,
Fig. 7. Sending station QoSCalc(SMS) – LTE.
the analysis section can be initialized with the analysis
button.
Analysis section. At the current state of development, this
section is only active for the receive station. It provides the
following information:
• “End-to-End Delivery Time for SMS” as a function
of the short message number,
• average “End-to-End Delivery Time for SMS”,
• “Completion Rate for SMS” for one measurement
scenario,
• the calculated “SMSQ-F”, based upon the user-
specified thresholds for the completion rate and de-
livery time and their weightings.
Finally, the new tool has to be tested in a real testing envi-
ronment to verify if it is working properly.
5. Measurement Environment and
Measurement Results
The measurement environment for the chosen measure-
ment scenario consists of two HTC 620G dual SIMs, each
plugged with a SIM card from Swisscom with LTE func-
tionality. Test location was 47◦12’25.2” N, 7◦33’17.3” E.
Unfortunately, the HTC 620G is not suitable for LTE test-
ing. Therefore, two Samsung Galaxy S6 Edge+ were been
included in the environment. Then, the following measure-
ment scenario was chosen.
7
Marcus Rompf and Tadeus Uhl
Fig. 8. Analysis section QoSCalc(SMS) – LTE.
Fig. 9. Results QoSCalc(SMS) – EDGE.
Fig. 10. Results QoSCalc(SMS) – UMTS.
Technology testing. Technology testing uses different
mobile network technologies. These are 4G (LTE), 3G
(UMTS) and 2G (EDGE). Each test consists of a set of
20 short messages sent at intervals of 10 s. The defined
thresholds were as follows:
• completion rate for all technologies: 1,
• delivery time: LTE – 1.3 s; UMTS – 4 s; EDGE –
5 s,
• the weightings are to be considered as equal.
8
The New Metric and the New Software Tool for Determining QoS in the Short Messages Service in Mobile Networks
Fig. 11. Results QoSCalc(SMS) – LTE.
Being successful in the measurement environment, the new
tool has delivered the results shown in Figs. 9 to 11.
EDGE results. These results show an average delivery
time of 5.06 s within an interval of [4.48; 5.73] and a com-
pletion rate of 1. With the chosen threshold, the SMSQ-F
is 0.99.
UMTS results. These results are showing an average de-
livery time of 3.83 s within an interval of [2.98; 4.28] and
a completion rate of 1. With the chosen threshold, the
SMSQ-F is 1.
LTE results. These results are showing an average delivery
time of 1.35 s within an interval of [0.82; 2.09] and a com-
pletion rate of 1. With the chosen threshold, the SMSQ-F
is 0.98.
Having analyzed all measurement results, one can conclude
that the new tool QoSCalc(SMS) is robust and suitable for
evaluating the QoS of SMS in different mobile networks,
such as LTE, UMTS and EDGE. The implemented metric
works properly and provides valuable results. It is possible
to adjust the thresholds to suit the testers’ needs, which
increases the flexibility of the tool considerably.
6. Summary and Outlook
This paper has been devoted to the subject of QoS mea-
surement within the SMS. It has focused on the creation of
a tool for measuring QoS in SMS for mobile network test-
ing. The new tool has been given the name QoSCalc(SMS).
It is based upon a new metric “SMSQ-F”, described in
Section 3, which takes both end-to-end delivery time for
SMS and completion rate for SMS into account. The im-
plemented tool has been tested in a real mobile network
environment. As far as the technology is concerned the
underlying measurement scenario and the respective results
display both a high degree of diversity, e.g. LTE, UMTS
and EDGE, and a high degree of flexibility regarding the
SMS quality factor, which is easily adjustable to meet the
end users’ needs when choosing the best service provider.
With that, the tool has proved its practical capabilities and
is ready to be used effectively in determining the quality of
the SMS in mobile networks.
In future work it would be very worthwhile to com-
pare the SMS service with OTT content messaging with
regard to reliability and especially availability. In ad-
dition, the tool should be improved to be more suitable
for test automation. Work in this direction is already in the
offing.
References
[1] Directives 2009/136/WE (Recom.EU L.337/11) [Online]. Available:
http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=
OJ:L:2009:337:0011:0036:en:PDF
[2] Official Journal of the European Communities, The European Par-
liament and Council, 07 03 2002 [Online]. Available:
http://www.ofcom.org.uk/static/archive/oftel/ind info/eu directives/
authorisation.pdf
[3] Official Journal of the European Communities, the European Parlia-
ment and Council, 12.07.2002 [Online]. Available:
http://www.dataprotection.ro/servlet/ViewDocument?id=201
[4] Official Journal of the European Union, The European Parliament
and Council, 27 10 2004 [Online]. Available: http://www.wipo.int/
wipolex/en/text.jsp?file id=199676
[5] ETSI TS 102 250-2 V2.2.1, “Speech and multimedia Transmission
Quality (STQ), QoS aspects for popular services in mobile networks,
Part 2: Definition of Quality of Service parameters and their com-
putation”, 2011.
[6] ETSI EG 202 057-2 V1.3.2, “Speech and multimedia Transmission
Quality (STQ). User related QoS parameter definitions and measure-
ments. Part 2: Voice telephony, Group 3 fax, modem data services,
and SMS”, 2011.
[7] Portio Research, “Mobile Messaging Futures 2014–2018” [Online].
Available: http://www.portioresearch.com/en/messaging-reports/
mobile-messaging-research/mobile-messaging-futures-2014-
2018.aspx
[8] SwissQual AG – A Rohde & Schwarz Company” [Online]. Avail-
able: http://www.swissqual.com/en
[9] Application Performance Index – Apdex Technical Specification Ver-
sion 1.1 [Online]. Available: http://www.apdex.org/specs.html
[10] T. Uhl and M. Rompf, “New Tool for Investigating QoS in the WWW
Service”, J. Telecommun. and Inform. Technol., no. 1, pp. 1–7,
2015.
9
Marcus Rompf and Tadeus Uhl
Marcus Rompf received his
M.Sc. in Information Technol-
ogy from the Kiel University
of Applied Sciences (Germany)
in 2016. Since 2011, he is a
voluntary Research Scientist at
the Institute of Communications
Technology, Flensburg Univer-
sity of Applied Sciences (Ger-
many), headed by Prof. Tadeus
Uhl in the field of performance
analysis of communication systems and evaluation of the
QoS and QoE in Triple Play services. Today he works as
Senior Research Engineer for Mobile Network Testing at
SwissQual AG (Zuchwil, Switzerland). His main activities
cover the following areas: Development of QoE and QoS
measurement applications for telecommunication services,
analysis of video, audio signals and data communication
in mobile networks, implementation of algorithms under
Linux and Android and support in design of measurement
frameworks in automated test systems.
E-mail: [email protected]
Flensburg University of Applied Sciences
Kanzlei st 91-93
D 24943 Flensburg, Germany
Tadeus Uhl received his M.Sc.
in Telecommunications from
the Academy of Technology
and Agriculture in Bydgoszcz,
Poland in 1975, Ph.D. from
Gdańsk University of Technol-
ogy, Gdańsk, Poland in 1982
and D.Sc. from University at
Dortmund, Germany in 1990.
Since 1992 he has worked as
Professor at the Institute of
Communications Technology, Flensburg University of Ap-
plied Sciences, Germany and in addition since 2013 as
Professor at the Institute of Transport Engineering and Eco-
nomics, Maritime University of Szczecin, Poland. His main
activities cover the following areas: traffic engineering, per-
formance analysis of communications systems, measure-
ment and evaluation of communication protocols, QoS and
QoE by Triple Play services, Ethernet and IP technology.
He is author or co-author of five books and some 130 pa-
pers on the subjects of LAN, WAN and NGN.
E-mail: [email protected]
Maritime University of Szczecin
Henryka Pobożnego st 11
PL 70-507 Szczecin, Poland
10